Query OpenTSDB without using aggregator functions like (sum , avg etc)

[anku...@gmail.com]

Hi,

I have been using OpenTSDB and storing time series data at certain intervals . My criteria is to get data from OpenTSDB only from the points where data is available. But going through the OpenTSDB documentation , I came to know that in query user have to provide an aggregate function and in the points data is not available , OpenTSDB linearly inter poles and calculate the data at that point and then perform aggregation and return the aggregated value . So while querying I am sometimes getting some ambiguous data.

Can I have data only where the data is available and avoid linear interpolation and aggregate functions.

Any help in this regard is highly appreciated. Thanks in advance.

--
Ankur.

[anku...@gmail.com]

Any one please reply.

[ManOLamancha]

On Tuesday, October 15, 2013 2:11:26 AM UTC-4, anku...@tcs.com wrote:

I have been using OpenTSDB and storing time series data at certain intervals . My criteria is to get data from OpenTSDB only from the points where data is available. But going through the OpenTSDB documentation , I came to know that in query user have to provide an aggregate function and in the points data is not available , OpenTSDB linearly inter poles and calculate the data at that point and then perform aggregation and return the aggregated value . So while querying I am sometimes getting some ambiguous data.

Can I have data only where the data is available and avoid linear interpolation and aggregate functions.

The easiest method is to simply request one time series at a time, that way you avoid interpolation altogether. This means you have to supply exact values for all of the tags that define the time series.

Or if you still want to aggregate multiple series, you can use the "zimsum", "mimmin" or "mimmax" aggregators with 2.0. zimsum supplies a 0 when a data point isn't found so sums should return what you expect. mimmin and mimmax return the min or max value for the current data type respectively.